Martini 3 Overestimates Lipid Scrambling by Membrane Proteins.

Abstract

Coarse-grained molecular dynamics simulations are widely used to study membrane proteins that mediate lipid translocation between bilayer leaflets, yet the reliability of current force fields for identifying and characterizing such lipid scrambling has not been systematically assessed. Here we find that around 20% of membrane proteins scramble lipids in Martini 3 simulations in the MemProtMD database, representing a 15-fold increase compared to the older Martini 2.2 force field. Martini 3 simulations overestimate lipid-scrambling rates by an average of 66-fold compared with Martini 2.2 and exceed experimentally measured rates. They also predict scrambling through closed scramblases and membrane proteins known not to scramble lipids, highlighting their limitations for studying lipid scrambling and lipid-protein interactions. Small refinements in the Martini 3 force-field parameters can remove spurious lipid scrambling, suggesting avenues for improving the accuracy of this model.

Abstract as supplied by the indexing source. © the authors / publisher.